Internet-based optical communications allow for transmissions of data with high speed and at large capacity. Also, recently, attempts have been made to use optical communications for large capacity computers and near distance communications.
In optical communications, data can be transmitted with high speed and at large capacity due to the possibility of simultaneous transmission of multiplexed optical signals along one optical path. Examples of core devices used in wavelength division multiplexing (WDM) for optical communications are light sources having a single wavelength, optical modulators, variable optical attenuators (VOAs), optical receivers, and optical switches. In particular, optical modulators, VOAs, and optical switches are frequently used in optical phase shifters. Accordingly, an optical phase shifter is very important for manufacturing an optical device. To manufacture an optical device with low costs, silicon is processed in a large surface using conventional techniques.
An example of a high speed optical device is a phase shifter using a variation in a refractive index obtained by a micro-electronic mechanical systems (MEMS) technique or a variation in a reflective index obtained by a thermal variation. However, this optical device has an operating speed of several KHZ to several MHZ, and thus cannot achieve data transmission rates higher than 1 Gbps. Accordingly, an electrical-optical system which can operate at high speed has to be used in this case.
FIG. 1 is a cross-sectional view illustrating a conventional p-i-n optical device, for example, an optical phase shifter or an optical attenuator.
Referring to FIG. 1, an undoped Si slab layer 20 is formed on a whole surface of a buried oxide layer 12 formed on a silicon on insulator (SOI) substrate 10. A portion of the undoped Si slab layer 20 is removed to form an optical waveguide 30 having a rib structure as illustrated in FIG. 1. An n+-ohmic layer 22a and a p+-ohmic layer 22b are arranged in the slab layer 20 on both sides of the optical waveguide 30 having a rib structure, thereby the slab layer 20 forming a p-i-n structure. The ohmic layers 22a and 22b contact a wiring 24 and the wiring 24 is respectively grounded or connected to a signal voltage Vsignal. The optical waveguide 30 is covered by a protection layer 32.
When a voltage in a forward direction is applied to the optical waveguide 30, charges 28 are formed in an optical region 26 of the optical waveguide 30, and thus the refractive index and the absorption coefficient of the optical waveguide 30 vary. However, since the dynamic characteristics of the p-i-n structured optical device are determined by the diffusion and recombination of minority carriers, the conventional p-i-n optical device has an operational speed of several hundreds of MHZ. Accordingly, it is difficult to use the optical device having the conventional p-i-n structure to achieve data transmission rates in the range of Gbps.
FIG. 2 is a cross-sectional view illustrating an optical device having a conventional metal-insulator-semiconductor (MIS) structure, for example, an optical phase shifter or an optical attenuator. The conventional optical device having the MIS structure is disclosed in detail in U.S. Pat. No. 6,801,676 to A. Liu et al.
Referring to FIG. 2, an n-type Si slab layer 50 is formed on a whole surface of a buried oxide layer 12 formed on an SOI substrate 10. A flat SiO2 gate insulating layer 60 having a width W is formed on a portion of the Si slab layer 50, and a current blocking layer 64 is formed on the slab layer 50 on both sides of the SiO2 gate insulating layer 60. The gate insulating layer 60 and the current blocking layer 64 are covered by a p-type Si upper layer or a polysilicon layer 62. The current blocking layer 64 is formed of SiO2 (refractive index=1.46) and thus has a lower refractive index than Si (refractive index=3.48) for blocking a current in a horizontal direction and guiding light. A pair of impurity regions 66 are formed on the polysilicon layer 62, and a second wiring 68 is formed on the impurity regions 66 for applying a signal voltage Vsignal. The n+-ohmic layer 52a or 52b in both upper sides of the slab layer 50 respectively contact first wirings 54 that are each grounded.
The impurity regions 66 are each a p+-type ohmic layer, and when a positive voltage is applied to the p+-type ohmic layer and a ground and negative potential is applied to the n+-type ohmic layer, a charged layer 58 of about 10 nm is formed on upper and lower surfaces of the gate insulating layer 60. The width of the gate insulating layer 60 is W, and the charged layer 58 is mainly included in an optical region 56. The charged layer 58 is formed of various charges, for example, holes 58a and electrons 58b, which accumulate on the opposite sides of the gate insulating layer 60.
FIG. 3 illustrates an optical mode distribution of the optical device of FIG. 2. The buried oxide layer 12 is formed to a thickness of about 3 μm, the n-type Si slab layer 50 is formed to a thickness of 1.43 μm, the flat SiO2 gate insulating layer 60 is formed to a thickness of 12 nm, and the polysilicon layer 62 and the current blocking layer 64 are formed to a thickness of about 1 μm. The width W of the gate insulating layer 60, that is, of the rib waveguide, is about 2.5 μm. In addition, the refractive index of Si is 3.48, and the refractive index of SiO2 is 1.46. The illustration of the optical mode distribution is shown for a horizontal direction and a vertical direction around the gate insulating layer 60.
As illustrated in FIG. 3, the intensity of the optical mode is the highest in the center portion of the gate insulating layer 60 in a horizontal direction. In detail, the intensity of the optical mode increases from both sides of the gate insulating layer 60 toward the center portion of the gate insulating layer 60. Also, regarding the vertical direction, the intensity of the optical mode of the Si slab layer 50 contacting the center portion of the gate insulating layer 60 is the highest.
FIG. 4A is a graph illustrating the optical mode distribution cut in a section along line 4A-4A (a perpendicular direction) according to the thickness of the gate insulating layer 60. FIG. 4B is a graph illustrating the optical mode distribution in a section along line 4B-4B (a horizontal direction) according to the thickness of the gate insulating layer. Line O-O is a virtual line illustrating the center portion of the gate insulating layer. The optical mode is expressed as a normalized optical profile, and the greater the profile, the higher the optical confinement factor of the charged layer which denotes the intensity of light.
Referring to FIGS. 4A and 4B, the smaller the thickness of the gate insulating layer 60, the higher the optical confinement factor of the charged layer 58. As the optical confinement factor increases, the phase shift of light increases and the propagation loss of light increases, thereby improving the performance of the optical device. Accordingly, the thickness of the gate insulating layer 60 is reduced to be thin enough to cause efficient optical phase shift and increased propagation loss of light. However, as the thickness of the gate insulating layer 60 is reduced, the cutoff frequency according to the increase capacitance decreases, and thus the dynamic characteristics of the optical device are degraded.
FIG. 5A is a perspective view illustrating a conventional gate-all-around (GAA) MIS structure, and FIG. 5B is a cross-sectional view of FIG. 5A. The MIS structure has been disclosed in ‘Conference on Laser & Electro-Optics (CLEO)’, pp. 110-112, 2005, by P. Dainesi et al.
Referring to FIGS. 5A and 5B, the GAA MIS structure has a channel layer 84, for example, a single crystal silicon, surrounded by a gate electrode, for example, a polysilicon, and a gate insulating layer 82, for example, a silicon oxide layer. The channel layer 84 has a shape having edge regions E at edges of the gate insulating layer 82 as illustrated in FIGS. 5A and 5B. An edge effect, which refers to the gathering of excessive amount of charges due to a voltage, occurs in the edge regions E of the gate insulating layer 82. The edge effect increases the optical phase modulation or optical attenuation effect (propagation loss of light) by the excessive amount of charges. Meanwhile, a field effect transistor (Fin-FET) MIS structure also shows a similar effect as the GAA MIS structure. However, it is difficult to apply a GAA or Fin-FET MIS structure to an optical device.